Change r34299 modified html4.css to allow media element controls to be placed anywhere on the media element, and updated the default positioning to be based on the bottom of the media element, not the top. This caused regressions in three layout-tests which were updated in r34376. One of these (media/video-controls-rendering.html) had the following diff: -layer at (8,746) size 320x18 backgroundClip at (0,0) size 785x762 clip at (0,0) size 785x762 outlineClip at (0,0) size 785x762 - RenderBlock (positioned) {DIV} at (0,224) size 320x16 +layer at (8,522) size 320x240 + RenderBlock (positioned) {DIV} at (0,0) size 320x240 The question is, why do we loose backgroundClip and outlineClip here, and does it matter? Visually, the actual results look the same as before, so we safely assumed we could update the results, but in case these two changes (backgroundClip and outlineClip) actually do something we need to go back and investigate further.
Assuming this was OK after all.